Multifractal Analysis of High-Frequency Temperature Time Series in the Urban Environment

Continuous monitoring systems have been regarded as a very useful tool to provide continuous high-frequency measurements of many parameters. Here, we analyze high-frequency time series of air temperature measurements, recorded every 10 min during 2003 in Athens (Greece) by an online monitoring system for the urban environment. We propose a set of time series analysis techniques, where missing data are well respected and information concerning the system’s dynamics is preserved. A power spectral density analysis is performed over time scales spanning from 10 min to several days. A scale-invariant behavior of the form E ( f ) ≈ f − β is revealed for scales below 9 h. Over this scaling range, we have performed structure functions analysis, and shown that air temperature data exhibit turbulent-like intermittent properties with multi-fractal statistics. The multifractal exponents obtained possess some similarities with passive scalar turbulence results. Although we illustrate the proposed approach using air temperature data, the method can be used as an efficient tool to analyse other environmental parameters monitored in urban environment